Measurement Issues: Differing Measurement Systems and Units
From an AEI report on seismic airguns, which contains footnotes and bibliography, detailing all references .
[DOWNLOAD REPORT (rtf)]
More surprising than the forgoing discussion of dB differences is the fact that there is no generally accepted measurement system for use in studies of underwater sound. Several systems are in use, each with its own advantages and relevance to the impacts of sound on animals; again, similar sounds result in significantly variable dB measurements, depending on which measuring system was used. Strangely, many literature surveys complain that all too often, studies do not sufficiently clarify which system is being used, making comparisons between studies difficult; in recent years, this difficulty has been more widely recognized, so that while there is still no agreed upon standard approach, at least researchers communicate more clearly which one they are using.
The three most common systems of measurement are:
- Peak levels / peak-to-peak, mean peak, or zero-to-peak (measured in units of dB re 1uPa2; though often by convention the 2 is omitted). This considers the change in amplitude (pressure) of a sound wave from the lowest to highest point on its waveform. It is relatively easy to measure, and may be especially relevant to concerns about direct physical damage to tissues, since it is the best reflection of the physical displacement likely to occur in tissue with the passing of a sound wave. In addition, peak measurements are less time-dependant than the following methods, which is appropriate for short-duration sounds such as airguns. However, especially as sound is measured at more distance from the source, reflection of the sound waves off the seas bottom and surface, along with differences in the speed at which sounds travel through different layers of the sea, create a situation where the received sounds are arriving along countless different pathways, making their waves interfere with each other. This interference can decrease or increase the intensity of the waves arriving at the receiver; accurately predicting the actual peak values is virtually impossible in many cases.
- Root-mean-squared or mean-squared-pressure levels (units of dB re 1uPa). This measures the total sound intensity, then divides it by the length of the signal. It is also useful biologically, because our perception of a sounds intensity takes place over time, not instantaneously. Acoustic power, intensity, and energy are proportional to the mean squared pressure. However, measurements based on this system are difficult, since it is not always easy to precisely identify when a sound starts and stops, especially at some distance from the source, where individual sound impulses are blurred and often can be confused with existing background sounds (at the lowest frequencies, headwaves traveling faster along the bottom of the ocean, also complicate the timing). This problem can tend to cause researchers to over-estimate the length of the sound, thereby under-measuring the RMS value.
- Measurements of the signals energy (units of dB re 1uPa2/s). Like RMS values, energy measurements do not depend so directly on the specific waveform of the received sound. After extensive comparisons of the many approaches to measurement and prediction, McCauley (2000) built on earlier work by Malme et al (1996) to develop a formula that has proven effective in measuring the energy of a signal; simply stated, it involves a correction of mean squared calculations to account for the difficulty in determining signal duration. He found that, compared to the above systems, measurements of what he termed ?Equivalent Energy? provided the most reliable indication of a signal’s intensity in a wide variety of situations (water depth, distance downrange, etc.).
While McCauleys Equivalent Energy approach proved to be effective at accurately measuring sound in a variety of situations, in order to provide more useful comparison to other systems, and to better estimate the effects that are best measured by the other systems, a conversion factor was obtained, based on direct measurements in the field. For an airgun array in the open ocean, RMS values averaged about 13dB higher than Equivalent Energy values, while peak-to-peak values averaged 28dB higher.
Note: Each of the above systems presents measurements based on a signal’s total broadband energy (generally using the 1/3 octave band standard to create the subsets which are summed or averaged); most commonly, these energy levels are plotted against distance, to show how the received level falls as it travels. At times, they are stated with reference to the range of frequencies being measured (generally the range within which the sound had significant energy). It is also common to present sound data as "sound pressure density spectra" levels. In this approach, the sound pressure or intensity is broken down and plotted separately in smaller frequency band chunks; the resulting graphs show the sound level (in units of dB re 1uPa2 /Hz) plotted across the frequency spectrum, from 1Hz to a relevant number of kHz. This approach is useful to analyze the tones at which a broadband sound is most powerful; it can also highlight the general bands within which animal sounds or anthropogenic sounds carry the most energy. As related to seismic airgun signals, sound pressure density spectra levels tend to highlight the ways that airgun sounds are especially intense in frequencies below 100Hz, while carrying continued but declining energy up to 1 or 2 kHz. McCauley (2000) and others often present density spectra data as well as broadband energy data, either by plotting the sound level at a given distance against the frequency spectrum, or by adding a color scheme that allows presentation of three variables: frequency spectrum, dB levels, and distance.